Re: detecting replication issues

Joseph Lorenzini <jaloren@xxxxxxxxx> · Fri, 24 Feb 2017 12:01:18 -0600

Hi Alessandro,
That will address the failover issue but it will not address configuring the glusterfs client to connect to the brick over TLS. I would be happy to be wrong. I was only able to get both by specifying that in the config file. What's curious is why the config file doesn't handle replication the same way as when its mounted with the mount command. I'd figure they should be the same.

Here's my config file if anyone is interested. Perhaps I don't have something set properly?

 volume gv0-client-0
     type protocol/client
     option ping-timeout 42
     option remote-host host1
     option remote-subvolume /data/glusterfs/gv0/brick1/brick
     option transport-type socket
     option transport.address-family inet
     option send-gids true
     option transport.socket.ssl-enabled on
 end-volume
 volume gv0-client-1
     type protocol/client
     option ping-timeout 42
     option remote-host host2
     option remote-subvolume /data/glusterfs/gv0/brick1/brick
     option transport-type socket
     option transport.address-family inet
     option send-gids true
     option transport.socket.ssl-enabled on
 end-volume
 volume gv0-client-2
     type protocol/client
     option ping-timeout 42
     option remote-host host3
     option remote-subvolume /data/glusterfs/gv0/brick1/brick
     option transport-type socket
     option transport.address-family inet
     option send-gids true
     option transport.socket.ssl-enabled on
 end-volume
 volume gv0-replicate-0
    type cluster/replicate
    subvolumes gv0-client-0 gv0-client-1 gv0-client-2
 end-volume

Joe 

On Fri, Feb 24, 2017 at 11:40 AM, Alessandro Briosi <ab1@xxxxxxxxxxx> wrote:

    Il 24/02/2017 14:50, Joseph Lorenzini
      ha scritto:

      1. I want the mount /etc/fstab to be able to fail over to any
        one of the three servers that I have. so if one server is down,
        the client can still mount from servers 2 and 3.

    backupvolfile-server option

    should do the work or use the config file.

    It's mentioned in the blog you linked...

    If you need more dynamic failover probably rrdns could be a
    solution.

      Alessandro

_______________________________________________
Gluster-users mailing list
Gluster-users@xxxxxxxxxxx
http://lists.gluster.org/mailman/listinfo/gluster-users